October 24, 2022
7 minute read
If the last three years have been all about the rise of retail media, the next three are going to be about measurement.
Amazon, in particular, is advancing data and analytics tools for consumer brands at a rapid pace, driving us closer to real time business management, a key theme in Stratably’s retail outlook.
The challenge for brands over the next three years will be to implement an organizational structure, tech stack and processes that make response times instantaneous rather than measured in weeks, months or, in some cases, years.
Last week, I connected with Andrew Hamada, Founder and CEO of Reason Automation, to explore how the Amazon platform is evolving through a data lens.
The following interview is most helpful to VPs of eCommerce, National Account Managers on Amazon, and Analytics leaders that want to understand:
- Data versus analytics
- Root causes that create data silos
- The biggest change in reporting this year
- What’s driving accelerating analytics innovation
- What data the most advanced brands are focusing on this year
Andrew has a unique vantage point into the world of Amazon data:
- He co-founded Reason Automation in 2020 after working at Amazon for seven years in a variety of roles. His company transforms the labor intensive, error prone process of extracting data out of Amazon’s system into an automated, reliable managed database service.
- Reason Automation serves over five hundred brands with over $5 billion in Amazon annual revenue, mostly in consumables and apparel, and has bootstrapped its way to profitability.
- Share of voice is a top data metric for brands
- Brands must move to the new Retail Analytics tool despite challenges
- Support specialists and engineering teams are steadily replacing vendor managers
- Accurate financial data on remittances and chargebacks can drive 50-100 bps of margin
- Faster feedback loops – decision-making that previously took months is shifting to real-time
Thanks Andrew for taking the time to educate me on the world of Amazon data. Let’s start with understanding how Reason Automation got started?
Our time at Amazon taught us how to structure enterprise-scale ecommerce data for analysis, how to create powerful analytical tools, and how to make data-driven decisions. We brought those skills to the other side of the table for manufacturers and agencies and were surprised to learn that the wealth of internal data is not made available programmatically to brands. Instead, you get the limited interfaces in VC, SC, and the Ad console, which can be taxing to work with.
Just charting an ASIN’s performance over time can require downloading multiple reports. When pulling data takes time, you do it less often—and anyone managing an Amazon business knows that speed and frequency matter. If ad performance fluctuates daily, can you afford to look at your business data weekly or monthly?
We created the technology powering Reason Automation to serve our own need: a fast SQL database of all Amazon sales, inventory, advertising, and other data. Once built, we knew others would use the database because the Amazon-adjacent industry is populated by people like us: ex-Amazon business users who are used to working with SQL and ETL. There was no product for us. Now we’ve recreated the experience of using the old “ASIN magic” tool: pulling data and writing queries directly from Excel. Then the same database can power your Tableau, Power BI, Looker and other analytics.
What's the difference between data and analytics in the context of how a brand should think about Reason Automation?
Data is the raw material from which insights can be derived. The mechanisms we use to derive insights from prepared data are analytics, the dashboards and reports and spreadsheets that most people think of when they think about using data. But those analytical tools are useless without clean, prepared, trustworthy input data.
There are hundreds of Amazon dashboards and analytics platforms but not all make that data programmatically available to customers without signing up for the entire platform cost. The subscription costs can be tough to justify if you just want to export the data into your ERP or omnichannel analytics platform anyway.
It sounds like the ideal client for you has a fairly good sense of what they want to build with Amazon data, they just don't have a good way to get to a good starting point with the data…is that right?
Exactly. Or they’ve resorted to highly manual processes, like having an analyst or virtual assistant download and aggregate reports from Amazon.
Who are you typically working with inside a consumer brand client?
Depending on the size of the brand it’s the Head of Ecommerce, Head of Amazon Sales, maybe the Director or VP of Operations—someone who feels the P&L pain of data inefficiency or missed analytical opportunity. If the brand has an established data practice, it’s a Senior Analyst, Director of BI/Analytics, or someone responsible for data governance.
What are the root causes towards silos when it comes to Amazon data?
For most companies it’s because they have existed longer than they’ve sold on Amazon and or Amazon does not account for most of their business. Imagine onboarding Amazon in the mid-2000’s when Walmart was still 30-40x their revenue (and not giving you programmatic sales data anyway). You must be content with whatever your vendor manager emails you - probably a PDF or PowerPoint.
Over time Amazon launches a marketplace, self-serve advertising, and other businesses that engage new aspects of your company. The data reaches your company in silos. They all roll up to different VPs and budgets, so data coordination is difficult even if you want it.
What's changing from a data perspective at Amazon this year?
Amazon’s technology release pace has been speeding up for the past 5 years but is accelerating dramatically this year. In the same year we saw expanded seller data and improved reports like Inventory Ledger, new brand analytics tools like Search Query Performance, a whole suite of vendor report APIs for the first time in Amazon’s history, dozens of new Advertising reports, Marketing Cloud, and more.
What do you attribute that to?
Amazon’s skyrocketing advertising business, which I don’t expect to slow. They have amazing privacy-related tailwinds like the death of the cookie and iOS’s privacy updates that make it harder for other advertisers to collect data. If you’re an Amazon customer, you must let them cookie you to sign in and shop, so they can track and attribute your shopping to your ad behavior. When Amazon shares this data, brands get ad + funnel analytics second only to their own DTC, and maybe not even then.
Why do you say “maybe not even then” as it pertains to Amazon’s data compared to DTC data?
Cross-platform attribution has always been fragile, but those iOS privacy changes mean that now iPhone users can click your ads and buy on your website without being tracked. So, it's very difficult to associate the ad metrics with commerce metrics. Amazon doesn't have that problem since most people are logged into their Amazon account when they shop.
What changes to reporting have occurred this quarter on Amazon that have been most impactful?
Without a doubt in my mind, the (re)launch of Retail Analytics for vendors in September. A complete overhaul of the vendor business reporting suite, the new Retail Analytics left many users cold. There were some undeniable benefits, like 3 years of daily sales history, consolidating all the sales diagnostic reports into one, and usability improvements to the forecasting report.
But at great cost. Lost buybox, a critical metric measuring the share of time your offer vs. competitors is presented to customers on a detail page, was deprecated without a replacement or explanation. Daily inventory data was deprecated, and weekly inventory history is limited to 6 weeks.
Although the old reports are still available, Amazon has made it clear that vendors should use the new Retail Analytics suite, and that vendor management & other internal teams are doing the same. Frustrating!
Brands get frustrated by Amazon changing reporting on what feels like a whim, with little to no warning in advance. Why do they do it that way?
A couple big reasons.
When I still worked there, Amazon preferred decisions to be made quickly, and that ambiguity be resolved through testing and data collection. Put those together and Amazon loves introducing changes at scale to get fast, direct user feedback and interaction data.
But Amazon also historically had a “two-pizza” concept for engineering teams, where a team should always be small enough to feed with two pizzas. While they’re obviously not strict about that anymore, the attitude lives on and Amazon services are often an amalgamation of 20+ child services and products maintained by separate teams.
Even if or when this changes, the architecture needs a lot of refactoring before unlocking consolidation benefits. This means that changes can be deployed somewhat haphazardly, often without the forenotice brands and other users expect.
Many brands I talk with describe challenges with different reports from Amazon not ticking and tying…why is that the case and is it cause for concern?
Mainly because of the same engineering diversity challenge I just mentioned: small teams each owning portions of a product, instead of a centrally planned monolith. Since different teams build things differently, this adds diversity where it isn’t wanted, like to report schemas, underlying data sources, etc.
That all gets exacerbated because Amazon is a global company. For example, different teams handle time zones and date cut-offs differently. In some cases, a report may be in Pacific Time (since that’s HQ), in other cases in GMT, in still others they will render in whatever time zone the user’s computer is in. A seller in Spain may have to bridge reports rendering in three different time zones.
It’s not ideal, but every company has these quirks. Have you ever tried to tie Google Analytics and Google Search Console data, or aggregate individual keywords to total page metrics? It’s just not going to work.
What type of Amazon data are the most advanced companies you work with seeking or paying attention to?
In general, search share of voice is a major focus. This tells a brand how often and how high their products rank for search terms they care about. Many companies consider this the most-important digital shelf metric, but there are many different ways of measuring it.
In terms of data we provide, it’s financial data. For vendors this means invoice and remittance data, chargebacks, and other financial optimizations that help ensure payments from Amazon are as large as you’re entitled to. After a point there’s more ROI engineering 50-100 bps of margin than further optimizing ads.
There's a lot of innovation happening on the analytics side of things. At the same time, 1P retail sales are decelerating into the mid-single digit range and I'm getting consistent feedback on VMs leaving and Amazon not backfilling. It feels like Amazon is pulling back on the retail organization from a people perspective. Do you agree, how do those things square - fast innovation but also pulling back?
The future of every Amazon organization is the same: more automation and self-service.
Through that lens, the faster innovation and pullback of human managers can not only coexist, but they’re also complementary.
Better data and reporting interfaces, more direct management tools like sponsored ads and self-service deals, and you eventually don’t need vendor managers—you need support specialists for the tools, and engineers to resolve escalations.
What's different about consumer brand organizations that really excel from a data/analytics perspective?
Data must talk. If you can change someone’s mind by showing them detailed market research proving them wrong, you’re more likely to invest in instrumentation & telemetry for your eCommerce business. You’ll expect data-driven arguments from your teams and peers, and always from yourself. You will let yourself be convinced by data when you might not be otherwise. Ambitious, driven people will notice this and make data part of their practice.
That’s really it. If people want to use data, they will push their leaders to make it more accessible, their leaders will push them to justify the expense, they will commit to larger goals, and hopefully succeed and create a flywheel of investment and belief in the analytics practice.
What does Reason Automation look like in three years?
Although it’s always tempting to go wider, our customers trust us because we’re so focused on providing high-quality Amazon data. With that in mind we’ll start 2023 focused on data accessibility improvements through new products (e.g., simplified schemas), documentation (e.g., tutorial videos), and services (e.g., diagnostic consulting).
Beyond that it’s hard to say, but our strength in data extraction and operations can easily be pointed at other retailers like Walmart, Target, Instacart, Shopify, Home Depot, etc. Once we’re confident we can keep up with Amazon’s accelerating release pace, we’ll consider broadening our horizons.